基于颜色分割和多特征融合的交通标志检测  

Traffic Sign Detection Based on Color Segmentation and Multi-features Fusion

作　　者：卢盛荣[1] 刘礼锋[1] 李翠华[1] 

机构地区：[1]厦门大学信息科学与技术学院,福建厦门361005

出　　处：《厦门大学学报（自然科学版）》2011年第4期685-689,共5页Journal of Xiamen University：Natural Science

基　　金：国防基础科研计划项目(B1420110155);国家重点基础研究发展计划(973)项目(2007CB311005);福建省教育厅A类项目(JA09230;JA09231)

摘　　要：采用了一种鲁棒的交通标志检测算法,该算法结合了基于颜色分割的粗定位过程和基于多特征融合的交通标志精确定位过程.粗定位利用交通标志的颜色特征,采用基于YIQ空间的颜色分割方法,获得图像中有可能包含交通标志的图像子区域;基于多特征融合的精确定位是采用梯度方向直方图(histogram of oriented gradient,HOG)及局域二值模式(local binary pattern,LBP)两种互补的特征,并利用支持向量机(support vector machinc,SVM)进行分类,得到交通标志的准确位置.实验表明该方法对亮度变化、视点变换、尺度变化及目标部分遮挡等情况具有很强的鲁棒性,并且查准率和查全率总体都比基于单特征的方法好.Traffic sign detection is important in intelligent transport system.In this paper,an efficient novel approach is proposed to achieve automatic traffic sign detection.The detection method combines color segmentation with learning based multi-features of traffic sign guided search.The rough location stage could obtain possible region of traffic sign using color segmentation based on YIQ space.The exact location stage searches traffic sign in these traffic sign possible regions based on multi-features fusion,we use histogram of oriented gradient（HOG） and local binary pattern（LBP） to classify by support vector machine（SVM）.Experimental results show that,the proposed approach can achieve robustness to illumination,scale,viewpoint change and even partial occlusion.The average detection rate and the false positive rate of our approach are better than the method based on one feature.

关 键 词：交通标志检测 颜色分割 梯度方向直方图 局域二值模式 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]